Home icon Chevron Chevron Blog Chevron News Chevron

PoC Released for Fortinet Vulnerability CVE-2025-3275

PoC Released for Fortinet Vulnerability CVE-2025-3275

May 23, 2025


FortinGuard Labs issued an advisory for CVE-2025-32756, a critical vulnerability affecting multiple Fortinet products. Just a day later, CVE-2025-32756 was added to CISA’s Known Exploited Vulnerabilities (KEV) catalog. Today researchers at Horizon3.ai released a proof-of-concept (PoC) demonstrating exploitation of the vulnerability. Their analysis focused on comparing the patched and unpatched versions of FortiMail.

CVE-2025-32756

According to the advisory, CVE-2025-32756 is a stack-based buffer overflow within the administrative API that allows unauthenticated remote code execution. The vulnerability stems from improper handling of user-supplied data in the AuthHash field of a cookie named APSCOOKIE.

Proof-of-concept (PoC) for CVE-2025-32756

Horizon3.ai started their research using the IOC provided in the advisory and confirmed admin.fe CGI binary is the exploitation target.

While the admin.fe binary remained identical across both versions, further investigation revealed that the issue resides in a shared library: libhttputil.so. A significant clue was the presence of cookie fields (Era, Payload, AuthHash) found only in this library.

     Ghidra Imports View(Source: Horizon3.ai)

Using tools like Ghidra, Horizon3.ai tracked the vulnerability to a function named cookieval_unwrap(), which decodes base64-encoded fields. The unpatched version lacks bounds checking for the decoded size of AuthHash, leading to a classic buffer overflow. The patched version introduces a check to ensure the input length is below a safe threshold (less than 30 bytes).

The overflow allows overwriting several stack variables, including saved registers and the return address. This enables precise control over program execution. By supplying a crafted base64-encoded string, the researchers were able to trigger the overflow and hijack execution.

Notably, the web server configuration and its use of mod_fcgid make this vulnerability easier to exploit reliably, as failed attempts are less likely to crash the application.

Conclusion

While Horizon3.ai did not publish a full exploit, their findings highlight the high exploitability of CVE-2025-32756 and underscore the urgent need to patch.

Given the nature of the vulnerability and the availability of a PoC, organizations using affected Fortinet products should update immediately and monitor for signs of compromise.

Source: hxxps[://]horizon3[.]ai/attack-research/attack-blogs/cve-2025-32756-low-rise-jeans-are-back-and-so-are-buffer-overflows/

Follow cybersecurity88 on X and LinkedIn for the latest cybersecurity news

Did you like the post? Share it in your media

Latest Articles

Featured

Bitget Hacked Through Zero-Day Vulnerability in Third-Party Security Product

September 30, 2026 · Chetna Sehgal

Bitget, one of the major cryptocurrency exchanges, has revealed new details about the cyberattack...

Read More >
Featured

AI Coding Agents Expose 13,000 Internal Screenshots on GitHub

September 30, 2026 · Chetna Sehgal

AI coding agents have unintentionally exposed more than 13,000 internal screenshots on public Git...

Read More >
Featured

Dutch Police Arrest Amsterdam Man in ShinyHunters Cybercrime Investigation

September 29, 2026 · Chetna Sehgal

Dutch authorities have arrested a 24-year-old man from Amsterdam as part of an investigation into...

Read More >
Featured

Vietnamese Man Charged in $16 Million Crypto “Pig Butchering” Scam

September 29, 2026 · Chetna Sehgal

A Vietnamese national has been charged in the United States for his alleged role in a major crypt...

Read More >
Featured

Microsoft Fixes Windows Bug That Broke File History Backups

September 24, 2026 · Chetna Sehgal

Microsoft has fixed a bug that caused the built-in File History backup feature to stop working on...

Read More >
Featured

OpenAI AI Agent Accessed Australian Medicare Portal, Raising New Cybersecurity Concerns

September 24, 2026 · Chetna Sehgal

OpenAI is facing scrutiny after one of its artificial intelligence agents gained unauthorized acc...

Read More >
Newsletter line